progs/parser-combinators/comb1.sc
author Christian Urban <christian.urban@kcl.ac.uk>
Tue, 04 Apr 2023 22:31:09 +0100
changeset 906 2bf1516d730f
parent 898 45a48c47dcca
child 919 53f08d873e09
permissions -rw-r--r--
updated

// Parser Combinators: Simple Version
//====================================
//
// Call with
//
//  amm comb1.sc

 
//  Note, in the lectures I did not show the type bound
//  using is: I => Seq[_], which means that the input 
//  type 'I' needs to be a sequence. 

abstract class Parser[I, T](using is: I => Seq[_])  {
  def parse(in: I): Set[(T, I)]  

  def parse_all(in: I) : Set[T] =
    for ((hd, tl) <- parse(in); 
        if is(tl).isEmpty) yield hd
}

// parser combinators

// alternative parser
class AltParser[I, T](p: => Parser[I, T], 
                      q: => Parser[I, T])(using I => Seq[_]) extends Parser[I, T] {
  def parse(in: I) = p.parse(in) ++ q.parse(in)   
}

// sequence parser
class SeqParser[I, T, S](p: => Parser[I, T], 
                         q: => Parser[I, S])(using I => Seq[_]) extends Parser[I, (T, S)] {
  def parse(in: I) = 
    for ((hd1, tl1) <- p.parse(in); 
         (hd2, tl2) <- q.parse(tl1)) yield ((hd1, hd2), tl2)
}

// map parser
class MapParser[I, T, S](p: => Parser[I, T], 
                         f: T => S)(using I => Seq[_]) extends Parser[I, S] {
  def parse(in: I) = for ((hd, tl) <- p.parse(in)) yield (f(hd), tl)
}



// an example of an atomic parser for characters
case class CharParser(c: Char) extends Parser[String, Char] {
  def parse(in: String) = 
    if (in != "" && in.head == c) Set((c, in.tail)) else Set()
}

CharParser('a').parse("abc")

// an atomic parser for parsing strings according to a regex
import scala.util.matching.Regex

case class RegexParser(reg: Regex) extends Parser[String, String] {
  def parse(in: String) = reg.findPrefixMatchOf(in) match {
    case None => Set()
    case Some(m) => Set((m.matched, m.after.toString))  
  }
}

// atomic parsers for numbers and "verbatim" strings 
val NumParser = RegexParser("[0-9]+".r)
def StrParser(s: String) = RegexParser(Regex.quote(s).r)

NumParser.parse("a123a123bc") 
StrParser("else").parse("elsethen")


// NumParserInt transforms a "string integer" into a propper Int
// (needs "new" because MapParser is not a case class)

val NumParserInt = new MapParser(NumParser, (s: String) => s.toInt)
NumParserInt.parse("123abc")

// the following string interpolation allows us to write 
// StrParser(_some_string_) more conveniently as 
//
// p"<_some_string_>" 

extension (sc: StringContext) 
  def p(args: Any*) = StrParser(sc.s(args:_*))


(p"else").parse("elsethen")           

// more convenient syntax for parser combinators
extension [I, T](p: Parser[I, T])(using I => Seq[_]) {
  def ||(q : => Parser[I, T]) = new AltParser[I, T](p, q)
  def ~[S] (q : => Parser[I, S]) = new SeqParser[I, T, S](p, q)
  def mapp[S](f: => T => S) = new MapParser[I, T, S](p, f)
}

def toU(s: String) = s.map(_.toUpper)

(p"ELSE").mapp(toU(_)).parse("ELSEifthen")  

// these implicits allow us to use an infix notation for
// sequences and alternatives; we also can write the usual
// map for a MapParser


// with this NumParserInt can now be written more conveniently
// as:

val NumParserInt2 = NumParser.map(_.toInt)


// A parser for palindromes (just returns them as string)
lazy val Pal : Parser[String, String] = {
   (p"a" ~ Pal ~ p"a").mapp{ case ((x, y), z) => s"$x$y$z" } || 
   (p"b" ~ Pal ~ p"b").mapp{ case ((x, y), z) => s"$x$y$z" } || 
    p"a" || p"b" || p""
}  

// examples
Pal.parse_all("abaaaba")
Pal.parse("abaaaba")

println("Palindrome: " + Pal.parse_all("abaaaba"))

// A parser for wellnested parentheses 
//
//   P ::= ( P ) P | epsilon
//
//   (transforms '(' -> '{' , ')' -> '}' )
lazy val P : Parser[String, String] = {
  (p"(" ~ P ~ p")" ~ P).mapp{ case (((_, x), _), y) => "{" + x + "}" + y } ||
  p""
}  

println(P.parse_all("(((()()))())"))
println(P.parse_all("(((()()))()))"))
println(P.parse_all(")("))
println(P.parse_all("()"))

// A parser for arithmetic expressions (Terms and Factors)

lazy val E: Parser[String, Int] = {
  (T ~ p"+" ~ E).map{ case ((x, _), z) => x + z } ||
  (T ~ p"-" ~ E).map{ case ((x, _), z) => x - z } || T }
lazy val T: Parser[String, Int] = {
  (F ~ p"*" ~ T).map{ case ((x, _), z) => x * z } || F }
lazy val F: Parser[String, Int] = {
  (p"(" ~ E ~ p")").map{ case ((_, y), _) => y } || NumParserInt }

println(E.parse_all("1+3+4"))
println(E.parse("1+3+4"))
println(E.parse_all("4*2+3"))
println(E.parse_all("4*(2+3)"))
println(E.parse_all("(4)*((2+3))"))
println(E.parse_all("4/2+3"))
println(E.parse("1 + 2 * 3"))
println(E.parse_all("(1+2)+3"))
println(E.parse_all("1+2+3"))


// with parser combinators (and other parsing algorithms)
// no left-recursion is allowed, otherwise the will loop

lazy val EL: Parser[String, Int] = 
  ((EL ~ p"+" ~ EL).map{ case ((x, y), z) => x + z} || 
   (EL ~ p"*" ~ EL).map{ case ((x, y), z) => x * z} ||
   (p"(" ~ EL ~ p")").map{ case ((x, y), z) => y} ||
   NumParserInt)

// this will run forever:
//println(EL.parse_all("1+2+3"))


// non-ambiguous vs ambiguous grammars

// ambiguous
lazy val S : Parser[String, String] =
  (p"1" ~ S ~ S).map{ case ((x, y), z) => x + y + z } || p""

//println(time(S.parse("1" * 10)))
//println(time(S.parse_all("1" * 10)))

// non-ambiguous
lazy val U : Parser[String, String] =
  (p"1" ~ U).map{ case (x, y) => x + y } || p""

//println(time(U.parse("1" * 10)))
//println(time(U.parse_all("1" * 10)))
println(U.parse("1" * 25))

U.parse("11")
U.parse("11111")
U.parse("11011")

U.parse_all("1" * 100)
U.parse_all("1" * 100 + "0")

// you can see the difference in second example
//S.parse_all("1" * 100)         // succeeds
//S.parse_all("1" * 100 + "0")   // fails


// A variant which counts how many 1s are parsed
lazy val UCount : Parser[String, Int] =
  (p"1" ~ UCount).map{ case (_, y) => y + 1 } || p"".map{ _ => 0 }

println(UCount.parse("11111"))
println(UCount.parse_all("11111"))

// Two single character parsers
lazy val One : Parser[String, String] = p"a"
lazy val Two : Parser[String, String] = p"b"

One.parse("a")
One.parse("aaa")

// note how the pairs nest to the left with sequence parsers
(One ~ One).parse("aaa")
(One ~ One ~ One).parse("aaa")
(One ~ One ~ One ~ One).parse("aaaa")

(One || Two).parse("aaa")



// a problem with the arithmetic expression parser: it 
// gets very slow with deeply nested parentheses

println("Runtime problem")
println(E.parse("1"))
println(E.parse("(1)"))
println(E.parse("((1))"))
println(E.parse("(((1)))"))
println(E.parse("((((1))))"))
//println(E.parse("((((((1))))))"))
//println(E.parse("(((((((1)))))))"))
//println(E.parse("((((((((1))))))))"))


// faster because of merge
lazy val E2: Parser[String, Int] = {
  (T2 ~ (p"+" || p"-") ~ E2).mapp[Int]{ case ((x, y), z) => if (y == "+") x + z else x - z} || T2 }
lazy val T2: Parser[String, Int] = {
  (F2 ~ p"*" ~ T2).mapp[Int]{ case ((x, _), z) => x * z } || F2 }
lazy val F2: Parser[String, Int] = {
  (p"(" ~ E2 ~ p")").mapp[Int]{ case ((_, y), _) => y } || NumParserInt }


println("Runtime problem")
println(E2.parse("1"))
println(E2.parse("(1)"))
println(E2.parse("((1))"))
println(E2.parse("(((1)))"))
println(E2.parse("((((1))))"))
//println(E2.parse("((((((1))))))"))
//println(E2.parse("(((((((1)))))))"))
//println(E2.parse("((((((((1))))))))"))